Module Quality Control & Assurance. Anthony Affolder University of California, Santa Barbara

Transcription

1 Anthony Affolder University of California, Santa Barbara 1

2 Outline Mechanical Quality Control & Assurance Gantry calibration & qualification Module quality Wire Bonding Qualification & Testing Qualification of pitch adaptors Wire bond pull tests Electrical Qualification & Testing Test stand types & part flow Test stand qualification & cross-calibration Module test & grading 2

3 Review of Production Network Si Sensors Companies CF plates Brussels Kapton cables&pins Bari,Aachen,CERN FE hybrid-asic Strasb.,Louv.,IC.. Pitch Adap Brussels Control and Distribution Center CERN Sensors Qualification Pisa-Perugia-Firenze Karlsruhe-Wien Strasbourg-Louvain CF Assembly Brussels, Pisa CERN-Pakistan Hybrid Assembly and Qualification CERN Focus of this talk Module Assembly Lyon-Brussels-Wien-Perugia-Bari-FNAL-UCSB Module Bonding and Testing Bari-Catania-Firenze-Padova-Pisa-Torino-FNAL-UCSB AAchen1&3-Karlsruhe-Strasbourg-Wien-Zurich Long Term Test Centers 3

4 General QC&A Philosophy How can process quality and uniformity be guaranteed in such a distributed system? 7 module assembly centers 13 module bonding and testing centers 9 sub-detector assembly and testing centers Require that equipment, software, and procedures used are as identical as possible Cross-calibrations are performed between sites in which standard candles (glass plates, hybrids, modules, etc.) are exchanged Require a high level of traceability of component/module flow and of testing results at all stages (Database) Global quality of production can be monitored Quick feed-back to production center, improving process quality and uniformity 4

5 Gantry centers Identical gantry systems assemble modules from components Bari and Perugia => TIB/TID Modules Brussels, Lyon and Wien => TEC Modules FNAL and UCSB =>TOB Modules Identical software used at all seven gantry cites TEC TIB TOB 5

6 Gantry Qualification All centers have qualified gantries through measurement of standard glass plates and production of multiple dummy modules The precision obtained is within our requirements 6

7 Module Mechanical Precision Before/after glue curing, fiducials of sensors, hybrid, and carbon fiber frame are measured Requirements on relative positions and angles between the sensors, the hybrid and the frame are made x and relative angle between sensors and between the frame and the sensors are most critical All modules (but 1) have passed these requirements On some modules, the measurements are repeated on an independent precision measuring machine All measurement are consistent in these cross checks. 7

8 Bonding Centers All centers are equipped with good bonding machines (mainly Delvotec and K&S) and are fully operational Bari,Catania, Firenze, Padova, Pisa, Torino TIB/TID Modules Aachen, Hamburg, Karlsruhe, Strasbourg, Wien, Zurich TEC Modules FNAL, UCSB TOB Modules Most of the centers have prior experience from other experiments Pitch adaptors and other test structures have been distributed between centers for bond quality testing 8

9 Bonding Qualification Each bonding center has been sent multiple bonding test pieces Wire bonds have their pull strengths, shapes, and break location studied at the center and at CERN All bonding centers have passed qualification with pull strengths greater than 9 grams Result of 1 st bond test ID# Centre CERN bond Self test CERN test comments 1 Torino 10.7 ± ± ± 0.9 Slight under deformation 2 Bari 11.6 ± ± ± 0.6 good 3 Catania 10.2 ± ± ± 0.4 Slight over deformation 4 Florence 11.1 ± ± ± 0.9 good 5 Karlsruhe 11.4 ± ± 0.5 ~10 ± 0.5 Very low loop:400µm 6 Strasbourg 11.6 ± ± ± 0.6 good, low loop:600µm 7 Vienna 10.3 ± ± ± 1.8 Lift-offs, large spread 8 Aachen 9.2 ± ± ± 2.0 Lift-offs, large spread 9 Padova ± ± ± 1.5 On Hughes, large spread 10 Padova ± ± ± 1.6 K&S, lift-offs, spread 11 Pisa ± ND 11.8 ± 0.7 Do destructive test 12 Pisa ± 0.4? Test new K&S 13 Zurich ± ± 1.1? 8.5 ± 0.8 Very high loop 14 Zurich ± ±? 14.6 ± 0.9 Amazing! 15 FNAL ± ± ± 1.1 Good but some lift-offs 16 UCSB ± ± 0.4 good 17 FNAL ± ± ± bad bond in CERN test, some lift-offs = corrected value based on CERN measured height and length in purple = some areas of improvement to work on 1 Alan Honma, CERN 9

10 Bonding QC&A During Production Pull strengths, bond deformation, and lift-off measurements are made on the test pad of pitch adaptor Every 50 th wire bond will be destructively pull tested and replaced on first ~20 module of each type at each test center Bond heads are regularly inspections and replaced when necessary A large number (>100) of hybrids and modules have been thermal cycled between 20 and 20 C with no wire bond breakage seen 10

11 Hybrid & Module Electrical Testing Hybrids and modules are tested during the various stages of module production Ensure that the performed operations do not introduce any or very few defects Currently require <2% faulty channels per module CERN, Louvain, Strasbourg, UCSB -> Hybrid Tests Bari/Catania, Firenze, Padova, Perugia, Pisa, Torino -> TIB/TID Modules Aachen, Brussels/Antwerper, Karlsruhe, Lyon, Louvain, Strasbourg, Wien, Zurich -> TEC Modules FNAL, UCSB -> TOB Modules A big effort of standardization in the hardware and software setups has been accomplished. Only ARCS and DAQ-based testing stand are foreseen to be used Both systems use the same data analysis and fault finding algorithms Data file and database outputs of the two systems are being standardized 11

12 ARCS Based Test Stands ARC FE And adaptor card ARCS - APV Readout Controller Software Purpose - Fast testing of hybrids and modules LED System DEPP LED Controller ARC Controllers Hybrid testing 28 test stands Module testing LED systems Pinhole/Open Tests DEPP HV supply Automated IV curves Pre-bonding Test (19 test stands) Post-bonding Test (15 test stands) 12

13 DAQ Based Test Stands DAQ system a PC based prototype of the real CMS tracker readout chain Purpose fast and burn-in testing of modules and sub-structures Single Module Test Stand (20 test stands) Substructure Assembly (10 test stands) and Burn-in (9 test stands) Module Burn-in (Wien box) (7 test stands) 13

14 Electronic Testing Cycle Quick test hybrid Gantry makes modules Quick test unbonded module Thermal cycle module Bonded module test Wirebond TEC/TOB Final pinhole test Assemble petals/rods Petal/rod burn-in 14

15 Hybrid Quick Tests Hybrids are tested prior to module assembly and bonding Ensure no damage has occurred during process Hybrids have had extensive QC & A prior to module assembly See F. Hartmann s talk The hybrid quick tests use the same systems (ARCS), algorithms, and fault finding requirements as the hybrid QC & A 15

16 Module Tests Module testing has matured greatly recently with the production of >50 modules of a given type. Minimum set of tests defined Fault finding algorithms are now tuned to maximize fault finding and fault type identification while minimizing false bad channel flagging Noise performance and shielding standardization has allowed for the same fault finding algorithms to work on the TIB, TEC & TOB Minimize the effects of external noise sources Results can be combined for the same module type measured at different sites in order to further refine testing Testing procedures are now almost automated Work to automate testing fault finding module grading database entry underway 16

17 Fault Finding Using ARCS (1) Noise Measurement Pulse Height Measurement (Using Calibration Pulse) Raw Noise vs. Channel Peak InvOff Pulse Height vs. Channel Dec InvOff Raw Noise [ADC Counts] sensor open Raw Noise CMS Noise Noisy 2 sensor open Pulse Peak [ADC Counts] Opens Pinhole Shorts 0.5 Pinholes Channel Channel Bad Channel Flags Bad Channel Flags 17

18 Fault Finding Using ARCS (2) Average Subtracted Peak Time (Calibration Pulse) Pinhole Test (Using LED System) Average Subtracted Peak Time (ns) Ave Sub Peak Time sensor open Peak InvOn 2 sensor open Pinholes h_risetimeaves Entries 1 Mean 2 RMS Channel Bad Channel Flags Calibration Injection Response Maximal Pulse Height Difference [ADC Counts] Difference between max and min pulse heights vs. channel LED Intensity Pinhole Peak InvOff Channel 18

19 Fault Finding Using ARCS (3) Test failures are correlated in order to diagnose fault type Open (1 or 2 sensor) Short Pinhole/Saturated Channel Mid-sensor opens Noisy Channels Faults are found >95% with correct fault type identified ~90% of the time Less than.1% of good channel flagged as faulty As more modules are built, fault finding criteria will be re-tuned to improve performance Database output of module testing is being finalized Similar tuning of fault finding underway for DAQ-based systems 19

20 Test Stand Cross-calibration All ARCS systems have had first iteration of cross-calibrations Modules are circulated between testing centers Multiple examples of common problems are added to each module Shorts (neighbors & next-to-neighbors) Opens (sensor-sensor & PA-sensor) Pinholes With new qualification standards, results nearly identical Final iteration of crosscalibrations are currently underway DAQ cross-calibration is forthcoming 20

21 Wien cold box Wien cold box cycles modules from 20 C to 20 C while reading out up to 10 modules DAQ Based System Modules cycled 1-3 days with ~4 cycles per day. TIB will cycle all modules TEC/TOB will cycle all modules until substructure burn-in available Torino Interlock Box VUTRI LV Distribution Peltier PS Multiplexer PAACB Electrometers Inside Wien Cold Box Backplane of Wien Cold Box HV Power Supply 21

22 Database Lyon Database Complete description of all parts and assembly Traces all movements of the parts between centres & keeps inventories Includes all test results performed at any centre Allows extract data for Controlling the production Tracing the anomalies Later calibration or slow control purposes Bias Current (na) Voltage Current(DB) Current(probing) Current (Bias Only) Current(Bonded) For example, I-V curve of a TOB module at various stages of production 22

23 Conclusions Module assembly is underway Assembled modules are very uniform and all meet specifications Wire bond strengths and shapes have been excellent Module electrical testing is semi-automatically finding the vast majority of faults with a very small false fault finding rate With the uniformity of results, the database allows for the tracking of global module quality Any systematic problems can be quickly identified and addressed at the production centers 23